Psychrophilic organisms are cold-adapted microorganisms that have overcome the constraints associated with living in permanently cold environments (Pikuta et al., 2007; Russell, 1990). Cold environments can induce molecular changes in an organism, including abnormalities such as low enzymatic activity, protein denaturation, and mis-folding of proteins (Russell, 1998). In order to overcome these deficiencies, several cold-adaptation mechanisms have been described, including cold-active enzymes, an unsaturated fatty-acid membrane, anti-freeze proteins, and sugar and/or alcohol compounds such as glycerol, trehalose, and sorbitol (Chattopadhyay, 2006; Feller, 2013; Feller and Gerday, 2003).

One of the most interesting cold-adaptation mechanisms is biofilm formation using extracellular polysaccharide substances (EPSs) that are found in the O-antigen component of lipopolysaccharides (LPS) and cell-bound polysaccharides (Bazaka et al., 2011). The EPSs play important temperature-buffering and cryoprotectant roles in cold-adapted microorganisms (Bazaka et al., 2011; Krembs et al., 2002). The psychrophilic bacterium Colwellia psychrerythraea strain 34H is a well-known producer of EPSs, which function as cryoprotectants (Marx et al., 2009). Increased production of EPSs is observed when samples are frozen from ?8 to ?14°C, which represents a strategy similar to that used by yeasts that increase the production of cryoprotectants (polyols and/or sugar alcohol) in response to osmotic stress (Marx et al., 2009; Shen et al., 1999). Recently, the molecular structure of capsular polysaccharides synthesized from C. psychrerythraea strain 34H was determined and characterized. Interestingly, it resembles other known antifreeze (Glyco) proteins and appears to inhibit ice recrystallization (Carillo et al., 2015).

C. psychrerythraea strain 34H was initially isolated from Arctic marine sediments (Huston et al., 2000). The temperature of growth ranges from ?12°C to 19°C and its optimal growth temperature is 8°C to 9°C (Wells and Deming, 2006). The results of genomic sequence analyses show that many gene products of C. psychrerythraea strain 34H play a role in the production and secretion of EPSs (Methe et al., 2005). These EPSs stabilize an extracellular protease, ColAP, against thermal denaturation (Huston et al., 2004). In addition, EPSs are involved in regulating intracellular biosynthetic reactions to proceed at extremely low temperatures; however, the external conditions that activate the production of EPSs remain unknown (Junge et al., 2006).

Gram-negative bacteria, including C. psychrerythraea strain 34H, contain either LPS or capsular polysaccharides (CPSs) in their outer membrane. Typical LPS is comprised of three covalently linked domains: lipid A, core oligosaccharide, and, in some bacteria, O-antigen. The core oligosaccharide is composed of inner and outer domains; the inner core contains 3-deoxy-d-manno-oct-2-ulosonic acid (Kdo), which is bound to lipid A, and L-glycero-d-manno heptose (heptose), which is linked to 3-deoxy-d-manno-oct-2-ulosonic acid (Brooke and Valvano, 1996; Nikaido, 2003; Raetz and Whitfield, 2002). Heptose is the connecting point for the variable outer-core saccharides that serve as the binding sites for O-antigen to produce complete LPS. Sedoheptulose 7-phosphate isomerase (GmhA), encoded by gmhA, catalyzes the conversion of sedoheptulose 7-phosphate (S7P) to d-glycero-d-mannoheptose 7-phosphate (M7P) in the first step of the heptose biosynthesis pathway (Kneidinger et al., 2001). Aberrant GmhA in Escherichia coli does not synthesize heptose and produces truncated LPS owing to the lack of a connecting point for more distal moieties (Brooke and Valvano, 1996).

The structure and catalytic mechanisms of GmhA from various species have been characterized because GmhA is considered an important target for antibiotic development (Harmer, 2010; Kim and Shin, 2009; Seetharaman et al., 2006; Taylor et al., 2008). In this study, we determined the crystal structure of GmhA from C. psychrerythraea strain 34H (CpsGmhA), which is critical for LPS biosynthesis and activity, and is potentially maintained at low temperatures for production of polysaccharides. In addition, sequence analysis reveals that CpsGmhA has high similarity with E. coli DnaA initiator-associating protein A (DiaA). Therefore, characterization of the structure of CpsGmhA may provide insight into the structural and functional correlations between GmhA and DiaA among microorganisms; additionally, comparisons of these proteins in mesophilic and psychrophilic organisms may help to elucidate the mechanism of low temperature-adapted enzymes.

Cloning, protein expression, and purification

The gene encoding GmhA protein (UniProtKB code: Q47VU0, amino acids 1?196) was amplified by polymerase chain reaction (PCR) from the genomic DNA of C. psychrerythraea strain 34H and subcloned into the pET 28a(+) vector (Novagen, USA). PCR amplification of the gene was performed using primers containing NdeI and XhoI restriction enzyme sites: forward primer 5′- CGATAACATATGCTAGAACAGATCAAAAAC-3′ and reverse primer 5′- CGATAACTCGAGTCATGAATCACCTTGCG GGAA-3′. E. coli BL21(DE3) cells were transformed with the cloned pET 28a(+) vector, which contains an N-terminal 6-histidine (His) tag and a thrombin cleavage site. The E. coli cells were grown at 37°C until the cultures reached an optical density of approximately 0.7 at 600 nm. Heterologous expression of the target protein was induced using 1 mM isopropyl-β-d-thiogalactopyranoside (Bioneer, Korea) at 25°C overnight, and the expressed protein was collected by centrifugation at 7000 rpm for 20 min. The cell pellet was resuspended in lysis buffer (50 mM sodium phosphate [pH 8.0], 300 mM NaCl, and 5 mM imidazole) containing phenylmethane sulfonyl fluoride and disrupted by sonication (Vibra-Cell VCX400; Sonics & Materials, Inc., USA).

Ni-NTA affinity resin (Clontech, USA) pre-equilibrated with lysis buffer was used to purify of the target protein. After washing the column and matrix with a ten-column volume of washing buffer (50 mM sodium phosphate [pH 8.0], 300 mM NaCl, and 35 mM imidazole), the His-tagged protein was eluted using a solution of the same composition but with 300 mM imidazole. Once the His tag was cleaved at 4°C overnight with thrombin (Sigma-Aldrich, USA), the resulting protein has an additional GSHM sequence at the N-terminus, an artifact of the cloning. The cleaved protein was loaded onto a Superdex-200 column (GE Healthcare, USA) equilibrated with 20 mM Tris-HCl (pH 8.0), 150 mM NaCl, and 1 mM dithiothreitol (DTT). The fractions containing CpsGmhA were collected and concentrated to 25.0 mg/ml for crystallization trials.

Crystallization, data collection, and structural determination

The purified CpsGmhA was subjected to an initial crystallization screen using the commercially available crystallization solution kits MCSG I?IV (Microlytic, USA), Wizard Classic I?IV (Emerald Bio, USA), and Classics and Classics II Suite (Qiagen, Germany), following the vapor diffusion method by mixing the protein (0.8 μl) and reservoir (0.8 μl) solutions. Small crystal granules were obtained within two or three days in 0.1 M Tris-HCl (pH 8.5), 0.2 M magnesium chloride, and 30% (v/v) PEG400 or 25% (w/v) PEG3350. The optimum crystallization condition was further refined. Flat hexagonal crystals suitable for X-ray data collection were formed in a hanging drop composed of 1 μl of CpsGmhA (25.0 mg/ml), 20 mM Tris-HCl, 150 mM NaCl, and 1 mM DTT, and 1 μl of reservoir solution in 0.1 M Tris-HCl (pH 7.2), 0.2 M magnesium chloride, and 19% (v/v) PEG400 against 500 μl reservoir. N-paratone was used as a cryoprotectant based on the results of a screening experiment. A native data set was collected at 0.97951 ? at the 5C beamline of the Pohang Light Source (PLS; Korea).

The data were recorded, integrated, and scaled using iMosflm software (Battye et al., 2011). In order to obtain the phase, molecular replacement (MR) was performed using the program MOLREP (Vagin and Teplyakov, 2009), using the homologous structure from E. coli DiaA (PDB databank accession number 2YVA; (Keyamura et al., 2007) as the model. A Matthews coefficient of 2.22 and 44.59% solvent indicated that one molecule was present in an asymmetric unit. After refinement up to 2.8 ? with REFMAC5 (Murshudov et al., 2011), the quality of the final structure was validated using the MOLPROBITY server (Chen et al., 2009) and SFCHECK from the CCP4 package (Vaguine et al., 1999), resulting in a final model with an R-factor of 21.0% (R_free 28.0%). The final refinement statistics are presented in Table 1. Figures and the surface and angle data were generated and calculated using PyMOL (DeLano, 2002).

Sequence analysis

The amino acid sequences of GmhA and DiaA from C. psychrerythraea strain 34H and other relevant species were obtained from the National Center for Biotechnology Information (NCBI) database (Table 2). Amino acid sequences were aligned using Clustal X (Chenna et al., 2003). Only well-aligned and conserved alignment sites were extracted from each alignment subset using Gblock (Castresana, 2000). Maximum likelihood (ML) analysis was carried out using RAxML Ver. 7.2.8 (Stamatakis, 2006). RAxML was run under the options GTRCAT and protein GAMMA, a complete random starting tree of 1000 bootstrap replicates (Pattengale et al., 2010).

Data deposition

The atomic coordinates and experimental structure have been deposited in the Protein Data Bank (accession codes 5BY2).

Overall structure

The GmhA from C. psychrerythraea strain 34H (CpsGmhA), including the additional serine-histidine residues located before the N-terminus as an artifact of cloning, was crystallized and the structure was determined by X-ray crystallography at a resolution of 2.8 ?. The monomeric structure is composed of five central parallel β-sheets surrounded by six α-helices (Fig. 1A). Most residues fit well in the electron density except for residues Tyr93, Gln94, Gln193, Gly194, Asp195, and Ser196. The final model was refined to R and R_free values of 21.0% and 28.0%, respectively (Table 1). The overall topology of CpsGmhA is very similar to that of the homologous structures from various species, with an average root mean square deviation (r.m.s.d.) of 1.3 ? in monomeric Cα (Harmer, 2010; Kim and Shin, 2009; Seetharaman et al., 2006; Taylor et al., 2008). Since one molecule of CpsGmhA was observed in the asymmetric unit, we performed a size exclusion chromatographic analysis. The result indicates that CpsGmhA also exists in tetrameric form in solution, which is in agreement with the results of previous studies (Harmer, 2010; Taylor et al., 2008) (Fig. 1B).

The bulk of helices (α1-α4 and α6) form a tetramer with a broad surface range. Half of the long N-terminal α1 helix interacts with the same region of the neighboring monomers via the hydrophobic residues Met1, Leu2, Ile5, Phe9, Ile13, Ile17, and Ile27, resulting in a four-helix frame around the tetramer (Figs. 1C and 1D). The four α1-helices in the tetramer form a hexagon tilted at an angle of 120°, which could be important for structural stabilization, because the four α1-helices surrounding the tetramer have broad contacting surfaces that serve as clamps on each of the two subunits (Fig. 1C). Extensive interactions with neighboring molecules support the order of the α3 helix, such that the overall shape has a closed conformation in comparison with homologous structures (see below) (Harmer, 2010; Taylor et al., 2008).

Active site

The active site of GmhA is mainly composed of four α-helices, which are each composed of two equivalent helices (α2?α6 and α2′?α6′) from two subunits. In addition, the conserved residues on the α3 helix and the three loops (β1?α2, α2?β2, and β3?α5) support the conformation of the active site (Fig. 2A). The E-Q-H residues, which are located at the α2, α6′, and α6 helices, respectively, are strongly involved in catalytic reactions. The E-QH residues are highly conserved in the GmhA family; however, the glutamate residue is replaced by glutamine at the equivalent position in CpsGmhA (Fig. 2A). Previous mutagenesis studies have shown no GmhA activity in EQ mutants (Harmer, 2010; Taylor et al., 2008), suggesting that CpsGmhA might be more closely related to DiaA this inference is supported by the bioinformatic analysis (see below).

The tetrameric structures of GmhA can be classified into two distinct conformations, open or closed, depending on whether an additional α3′ helix is present in the central region. The closed conformation may allow for the ingress and egress of a substrate and product in solution due to the extension of the solvent accessible area of the active site (Harmer, 2010). Since the conformation of CpsGmhA is closed, we compared the positions of the catalytic residues to those in other GmhAs with closed conformation from Burkholderia pseudomallei and Campylobacter jejuni (Fig. 2B). Most residues are oriented in a similar manner, except for the Glu63 residue of C. jejuni GmhA (CjGmhA), which is equivalent to the Gln62 residue in CpsGmhA. However, in comparison with those of the E. coli GmhA (EcGmhA), conformations of the catalytic residues are not well overlaid (Fig. 2C). In particular, the conformation of the Asp94 residue, which is equivalent to Asp92 in CpsGmhA, is quite different in that it faces the opposite side to Asp92. In addition, the positions of the Glu65 and Gln172 residues (equivalent to Gln62 and Gln172 of CpsGmhA, respectively) are different. Therefore, the conformations of the catalytic residues are affected by whether or not the overall structure is an open or closed conformation.

Comparison with homologous GmhA structures

Conformational differences between the open and closed may be because of the binding of the product to the active site. For example, the open-conformation structure of PaGmhA is converted to a closed structure by M7P binding (Taylor et al., 2008) (Fig. 3A). M7P binding drives the α4 helix to move away from the central cavity, resulting in a disordered α3 helix as a result of its interaction with the α2 of the other subunit. However, the B. pseudomallei and E. coli GmhA structures are not changed upon product or substrate binding (Harmer, 2010; Taylor et al., 2008) (Figs. 3B and 3C). In addition, the open conformations of EcGmhA and PaGmhA are slightly different in that the α2 and α6 helices in EcGmhA are tilted outwards at 25° and 15°, respectively (Fig. 3D). Therefore, the two pairs of subunits in EcGmhA are not symmetric, and the three different GmhA structures exist in two open states and one closed state (Fig. 3D). Taken together, no correlations are currently evident between the open and closed conformation and the structures.

Similarity between CpsGmhA and EcDiaA

We searched for structural homologs of CpsGmhA using the DALI server system and found that DiaA from E. coil (EcDiaA) has high similarity to GmhA, with a Z-score of 32.7 (Holm and Rosenstrom, 2010). DiaA is essential for the accurate initiation and precise timing of chromosomal replication during the cell cycle (Ishida et al., 2004). Thus, DiaA directly and specifically interacts with DnaA independent of nucleotides or DNA (Ishida et al., 2004; Keyamura et al., 2007). Moreover, the structure of DiaA shows similarity with the DnaA-binding protein HobA from Helicobacter pylori (HpHobA) (Natrajan et al., 2007). Based on this structural information, we compared the protein sequences and found that the catalytic glutamate residues are variable, with Gln (CpsGmhA), Ser (EcDiaA), and Asn (HpHobA) being observed at catalytic sites (Fig. 4A). In addition, the other catalytic residues are varied and HobA has no identical residues. Given the difference in the catalytic residues of HpHobA, we compared the conformations of the catalytic residues among CpsGmhA, EcGmhA, and EcDiaA (Fig. 4B). The conformation of the catalytic residues between CpsGmhA and EcDiaA is much more similar than that of other pairs of compared proteins because the two residues Glu65 and Gln172 are differently oriented. This result indicates that CpsGmhA and EcDiaA are closely related.

Next, we compared the structural similarity among the four structures. Structural comparisons show that CpsGmhA has higher similarity to EcDiaA with a r.m.s.d of 0.7 ?, but relatively low similarity to both EcGmhA and HpHobA with an r.m.s.d of 1.7 ? and 3.7 ?, respectively. Thus, the overall tetrameric structures of CpsGmhA and EcDiaA are completely overlaid (Fig. 4C). To investigate the sequence-based relationship between CpsGmhA and CpsDiaA, we performed a phylogenetic analysis of GmhA and DiaA (Fig. 5). GmhA and DiaA could be divided into two distinct groups. However, we found that GmhA and DiaA from C. psychrerythraea and Pseudomonas aeruginosa are not clearly separated. Indeed, GmhA and DiaA from the two species showed over 98% similarity because they differ in only two or three residues. This result indicates that PaDiaA could be classified as an isotype of PaGmhA and CpsGmhA as an isotype of CpsDiaA. Taken together, the structural and sequence similarities of CpsGmhA, CpsDiaA and EcDiaA may provide insight into the functional role of GmhA as an isotype of DiaA in C. psychrerythraea.

To date, eight crystal structures of GmhA have been reported from various mesophilic species (Harmer, 2010; Kim and Shin, 2009; Seetharaman et al., 2006; Taylor et al., 2008). The overall structures are divided into two forms: open and closed. The conformations of the apo-structures are also variable in that the GmhAs of E. coli, Vibrio cholera, and P. aeruginosa are open, whereas those of B. pseudomallei, C. jejuni, and C. psychrerythraea are closed. The overall structure of psychrophilic CpsGmhA is very similar to the homologous structures of the closed forms. Previous studies of GmhA crystal structures demonstrate that either two or four molecules are contained in the asymmetric unit. However, only one molecule of CpsGmhA was found in the asymmetric unit. The biologically reliable tetrameric structure was revealed by the C222 symmetry operation. Additional analysis of oligomerization using the program PDBePISA (v1.47) further indicated that biological assembly of the protein results in a tetramer (Krissinel and Henrick, 2007).

The reaction mechanisms of GmhA have been proposed by complex structures with either product or substrate (Harmer, 2010; Taylor et al., 2008). To date, there have been three GmhA complex structures reported in either substrate form or product-bound form. The substrate S7P is only present in the EcGmhA structure as an open conformation, and the product M7P is found bound in closed structures from P. aeruginosa and B. pseudomallei (Harmer, 2010; Taylor et al., 2008). Moreover, only one subunit of the tetrameric EcGmhA contains the substrate S7P, whereas BpGmhA and PaGmhA both have M7P products in each of the four subunits (Fig. 4C). Thus, the two subunits are not symmetrical about the vertical axis, as is the case with other GmhA structures. This may indicate that a local conformational change occurs upon substrate binding. However, there was no difference in symmetry between the protein and its apo-structure. On the other hand, it is also likely that a change in only one molecule of the substrate would not induce an overall conformational change.

As shown in the previous section, only CpsGmhA contains a catalytic glutamine, rather than a glutamate residue at the equivalent position (Gln62) in the active site. Therefore, CpsGmhA may not have isomerase activity, based on mutagenesis data reported previously (Harmer, 2010; Taylor et al., 2008). Furthermore, based on sequence identity, CpsGmhA is more closely related to EcDiaA (62.2% identical) than to EcGmhA (37% identical) or HpHobA (13.9% identical). Interestingly, CpsGmhA has the two signature residues, Leu190 and Phe191, in common with those of EcDiaA, which are critical for DnaA binding, whereas EcGmhA does not contain these residues (Fig. 5A) (Keyamura et al., 2007). In addition, there are only three residue discrepancies between CpsGmhA and CpsDiaA, and therefore no significant functional differences would be expected. Since a functional study of CpsGmhA has not yet been done, we propose that the CpsGmhA protein could be a putative isotype of CpsDiaA. Taken together, further investigation of CpsGmhA is required to confirm the functional similarity with CpsDiaA.